How can I add my custom RAG embedded Claude model with composer?

Discussion
1

Hi folks –

Apologies if this is documented somewhere but I’m having trouble figuring out how to set my custom RAG model as the one Composer uses.

My setup includes a pipeline that enhances Claude’s responses by retrieving relevant context from a Pinecone database of embeddings (documentation and code snippets for a specific library). When prompted, the pipeline retrieves and appends this context to the query before sending it to Claude via the Messages API. Currently just doing this with a py script but obviously getting the right context into the prompt and then feeding it back to cursor to build is challenging.

I assumed that I’d be able to somehow set this “modified Claude” as the model that composer uses in order to get more refined suggestions than base Sonnet. I’m sure I’m missing something but I haven’t been able to figure out how to set this up…

→ Is anyone able to point me at reference material that would help me understand how to set my custom model in composer, or offer any general tips?

Any guidance would be greatly appreciated, thanks in advance!

2

Hey, unfortunately, Cursor does not have support for users adding their own RAG pipeline, While you may be able to implement this into your own custom API, which you can use in the settings, the default context process cannot be changed or disabled for Cursor AI queries.